{
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\begin{tabular}{l*{3}{c}}
\hline\hline
            &\multicolumn{1}{c}{(1)}&\multicolumn{1}{c}{(2)}&\multicolumn{1}{c}{(3)}\\
            &\multicolumn{1}{c}{2014-2019}&\multicolumn{1}{c}{2009-2019}&\multicolumn{1}{c}{2009-2019}\\
\hline
2019 election&      -6.084\sym{***}&      -12.89\sym{***}&      -20.29\sym{***}\\
            &    (0.0631)         &    (0.0690)         &    (0.0741)         \\
[1em]
EP election in 2019&      0.0208         &      -1.043\sym{***}&       0.193\sym{**} \\
            &    (0.0715)         &    (0.0781)         &    (0.0754)         \\
[1em]
Ciutadella in 2019&       1.130         &       0.375         &       0.754         \\
            &     (0.693)         &     (0.589)         &     (0.589)         \\
[1em]
EP 2019 x Ciutadella&      -3.126\sym{***}&      -2.439\sym{**} &      -3.675\sym{***}\\
            &     (1.122)         &     (1.008)         &     (1.008)         \\
[1em]
2014-2015 election&                     &                     &      -14.04\sym{***}\\
            &                     &                     &    (0.0366)         \\
\hline
\(N\)       &      117934         &      204069         &      204069         \\
\hline\hline
\multicolumn{4}{l}{\footnotesize Standard errors in parentheses}\\
\multicolumn{4}{l}{\footnotesize All models include voting station * election type fixed effects}\\
\multicolumn{4}{l}{\footnotesize Standard errors are clustered by voting station * election type}\\
\multicolumn{4}{l}{\footnotesize \sym{*} \(p<0.10\), \sym{**} \(p<0.05\), \sym{***} \(p<0.01\)}\\
\end{tabular}
}
